Jump on the LOD Bus

Get on the LOD bus

Hopefully at this point you're thinking

If you're not thinking this, either you weren't paying attention or we failed ;-)

The LOD cloud

Basic principles of Linked Data

TimBL has delivered unto us the four commandments of Linked Data

  1. Use URIs as names for things

  2. Use HTTP URIs so that people can look up those names.

  3. When someone looks up a URI provide useful information, using the standards (RDF, SPARQL)

  4. Include links to other URIs so that they can discover more things.

this encourages the re-use of your data in wonderful and unexpected ways

6 Steps to publishing LD

Tom Heath from Talis has outlined the following steps

  1. Understand the principles
  2. Understand your data
  3. Choose URIs for Things in your data
  4. Setup your infrastructure
  5. Link to other data sets
  6. Describe and publicize your data

Understanding the principles

Hopefully we've covered this more or less

It really is all about applying TimBL's 4 rules to encourage re-use of data

for now let's just remember that...

There are plenty of manifestos and guides to read linkeddata.org is a good place to start

6 Steps to publishing LD

Cool, we understand the principles

  1. Understand the principles
  2. Understand your data
  3. Choose URIs for Things in your data
  4. Setup your infrastructure
  5. Link to other data sets
  6. Describe and publicize your data

Our example

musopen.org

Musopen contains a bunch of public domain classical music

We did a quick and dirty LD publishing of this data for our ISMIR paper on the MuSim similarity ontology

come see the talk tomorrow morning :-)

What's in your data?

Next we have a good think about what are the things in our data we want to publish

Think about vocab

Try to avoid re-inventing ontologies - have a look around and see what might work

Hacking the ontology

MO is great but maybe we want some shortcuts

MO modelling of a classical music work

Hacking the ontology

We use open.vocab to create some ad-hoc terms

MO modelling of a classical music work

We can define these terms in OV

open.vocab.org screen shot

Note this is abuse of MO

Yves is horrified

We're short-circuiting FRBR

This will shock and appall Yves

But anyone can say anything on the Web of Data

if you're gonna hack, hack it responsibly using open.vocab - this way all our terms dereference and we still obey rule 3

6 Steps to publishing LD

Cool, we understand our data

  1. Understand the principles
  2. Understand your data
  3. Choose URIs for Things in your data
  4. Setup your infrastructure
  5. Link to other data sets
  6. Describe and publicize your data

Let's mint some URIs

choosing URIs for your data is arguably the most important step

Of course we'll use HTTP URIs

in case you weren't sure



and remember cool URIs don't change

Mind the namespace

Use a namespace that you control

Abstract away from implementation details

try to keep the URIs free of clutter such as file extensions

Hash or Slash

Decide whether you want hash or slash URIs

This is really a matter of preference but probably better to stick with one or the other

Note if you have a very small dataset, the hash URI gives you the option of publishing a flat file (which is really easy)

Common URI patterns

http://dbpedia.org/resource/New_York_City← Thing
http://dbpedia.org/data/New_York_City← RDF data
http://dbpedia.org/page/New_York_City← HTML page
--
http://revyu.com/people/tom← Thing
http://revyu.com/people/tom/about/rdf← RDF data
http://revyu.com/people/tom/about/html← HTML page
--
http://www.bbc.co.uk/music/artists/db4624cf#artist← Thing
http://www.bbc.co.uk/music/artists/db4624cf.rdf← RDF data
http://www.bbc.co.uk/music/artists/db4624cf.html← HTML page

note these URIs have been cut short to fit on this slide

6 Steps to publishing LD

Cool, we've picked out some cool URIs

  1. Understand the principles
  2. Understand your data
  3. Choose URIs for Things in your data
  4. Setup your infrastructure
  5. Link to other data sets
  6. Describe and publicize your data

a quick note on RDFa

RDFa is RDF embedded in HTML

RDFa is an increasingly popular way to publish Linked Data

Back to our infrastructure

Way of the hash # (no conneg)

If you have a rather small and static dataset (like an ontology)

and you've decided to use a hash URI system (or slash and mod_rewrite)

it's easy!

Throw that file on your web server and you're good to go

now only one HTTP request is needed to grab all your data

(good if it's small, bad if it's large)

Way of the slash and the 303

Linked Data content negotiation can be a bit tricky (ask the BBC guys)

but don't worry


There's plenty of tools already out there to help

we'll discuss a few of them shortly...

First, setup our triple store first, then worry about the rest

Triple store are RDF databases

There are several really good open-source triple stores

We're using 4 Store

In our example we're using 4Store

Our workflow is rather simple:

  1. Parse the original data
  2. Model using our ontologies
  3. Write an RDF serialization to a file
  4. Load the RDF file into the store

Some more details

We use various Python scripts to generate various bits of the RDF

We use Sonic Annotator to generate audio features RDF

4Store gives us several options loading the RDF including a command line utility


	$ 4s-import mytriplestore /path/to/myfile.rdf

Apache + 4Store and we're almost there!

but...

Oh dear!

when we look up http://classical.catfishsmooth.net/resource/track/842 we get:

404 not found

that is not cool

We can do dereferencing

We've picked out cool URIs and just dumped them in our 4Store

but they don't dereference

don't worry, Pubby is a good solution

it's a Java-based server app that pulls RDF descriptions from a SPARQL endpoint and handles content negotiation in a nice Linked Data fashion

caveat: Pubby doesn't actually work with 4Store "out of the box" we had to do some hacking because 4Store can't handle describe queries (yet)

It's working

dereferencing works

get RDF with curl:


$ curl -I -H "Accept: application/rdf+xml" \
	http://classical.catfishsmooth.net/resource/track/842

6 Steps to publishing LD

Cool, we've got our infrastructure

  1. Understand the principles
  2. Understand your data
  3. Choose URIs for Things in your data
  4. Setup your infrastructure
  5. Link to other data sets
  6. Describe and publicize your data

Linking to music resources

When we're talking about music and Linked Data we're almost always talking about MusicBrainz

they provide URIs for artists, tracks, and albums

If your dataset deals with any of these things, you will want to create owl:sameAs links to Musicbrainz URIs in your dataset


<http://classical.catfishsmooth.net/resource/artist/123> 
		owl:sameAs 
			<http://musicbrainz.org/artist/123> .

GNAT and gnatlib

luckily there are some tools to help

6 Steps to publishing LD

Now we've just got to publicize our data

  1. Understand the principles
  2. Understand your data
  3. Choose URIs for Things in your data
  4. Setup your infrastructure
  5. Link to other data sets
  6. Describe and publicize your data

Publicize

Help others discover and index your data

Apply a license or waiver to your data set

now we've got more audio features stuff